|Fig. 1: Layers of soil. (Source: Wikimedia Commons)|
Radiocarbon dating of soils has always been a tricky problem. Since organic matter is continually being introduced into the soil, the measured age of soil organic matter has always tended to underestimate the true age of the soil.  Historically, a solution has been to separate the portion of the soil sample that has a greater age than the remaining fractions and use it to calculate estimate the soil age, but recently more advanced methods of soil dating have appeared, which give better estimations.
Carbon exists in the most part in the isotope C-12, but has a radioactive isotope, C-14, with a half-life of 5570 years. All terrestrial organisms use carbon dioxide in the atmosphere as a source of carbon, thus there is a constant exchange of C-14 with the atmosphere. When the organism dies, however, the exchange is stopped, and the C-14/C-12 starts decreasing with the radioactive decay of the C-14 present in the dead organism.  Thus, the amount of C-14 present in the soil can give us an estimate of its age.
Since the rate of radioactive decay is proportional to the number of radioactive atoms present, it is unnecessary to measure the amount of C-14 present in the soil sample. One need only measure the radioactivity per unit mass of carbon. Two systematic errors hamper the precision of radiocarbon dating: statistical errors and the fact that the atmospheric C-14/C-12 ratio is only approximate. The latter is due mainly to the temporal variations of cosmic radiation, the rise of stable carbon isotopes in the atmosphere due to increased consumption of fossil organic fuels (known as the Suess effect) and radioactivity caused by thermonuclear testing.
In order to minimize the amount of new carbon in the soil, the soil sample has to be liberated from coarse and fresh organic material, such as leaf and root tissue. Free carbonates in the soil are eliminated by treatment with hydrochlroic acid. The remaining material is then dried and burned to CO2, and the activity can then be measured by gas proportional counters or by liquid scintillation spectrometers. 
Taking the entire sample and measuring its radioactivity amounts to considering the entire sample to be of the same age, which entails ignoring the fact that organic material is continually added to the soil. This thus provides only a lower bound on the age of the soil. In order to improve the estimate, one might separate the sample into smaller fractions, thus the oldest fraction would be a lower bound of the soil age, giving a better estimate. 
A more advanced method of preparation, called "Turin's method", used for separating soil organics for dating, is detailed by Orlova and Panychev.  The organic matter is divided into five fractions: free humic acids, humic acids bound with Ca and R2O3 mobile species, humic acids bound with R2O3 stable forms, humin and soil remnant. First, sodium hydroxide is added to a dried sample, then clay particles are precipitated by sodium sulfate and one day later the solution is precipitated by the addition of sulfuric acid. The humic acids are then separated by repeated treatments by alkali in order to produce benzene, which is then used for dating.
One of the main problems with this method of soil radiocarbon dating is the presence of a steady state, beyond which 14C dating will yield no useful information regarding the age of the soil.  In order to understand when and in which situations such steady states occur, Orlova and Panychev studied soil samples from various parts of the former USSR. They concluded that 14C dates are valid in alluvial and flood deposits because of the relatively quick soil burial and thick overlying sediments which remove the buried soil from the zone of penetration of roots. The estimation is less accurate in loess deposits, in which the soil system remains open for a relatively long period. Fig. 1 shows how soils can differ in character between topsoil and buried soil.
Another method of tackling soil dating has been suggested by O'Brien and Stout.  They observe that the years preceding 1966 saw a sharp increase in the atmospheric concentration of radiocarbon. By studying the profiles of radiocarbon in the soil with respect to the depth, they came to the conclusion that the downward movement of this radiocarbon proceeds via a diffusion mechanism, and the depth of the diffusion is inversely proportional to the time squared. This model of diffusion allows for a much easier dating of buried soil. Given a "marker", for example a known volcanic eruption at a certain time in the past, by studying how much the volcanic soil has diffused into the ground, one should be able to date the soil using the diffusion method.
The above methods are only able to date soil approximately. Newer and better methods are being researched in order to decrease the errors in the estimations, and more sophisticated models have been proposed. As more data becomes available and models become more refined, one day we may be able to date soils to the same precision as fossils.
© Anrong Lin. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 Y. Wang, R. Amundson and S. Trumbore, "Radiocarbon Dating of Soil Organic Matter," Quaternary Res. 45, 282 (1996).
 H. W. Scharpenseel and H. Schiffmann, "Soil Radiocarbon Analysis and Soil Dating," Geophys. Surv. 3, 143 (1977).
 L. A. Orlova and V. A. Panychev, "The Reliability of Radiocarbon Dating Buried Soils," Radiocarbon 35, 369 (1993).
 B. J. O'Brien and J. D. Stout, "Movement and Turnover of Soil Organic Matter as Indicated by Carbon Isotrope Measurements," Soil Biol. Biochem. 10, 309 (1977).